[c-nsp] MRTG a L2TP VPDN Looback Interface...again

[Oliver Boehmer (oboehmer)]

Mark Tohill <> wrote on Wednesday, January 25, 2006 6:12 PM:


> Resurrecting a previous thread, we need to be able to measure BW on
> GigE interface to our upstream provider.
> 
> The problem is that L2TP tunnels terminating on Loopbacks  providing
> customer traffic can't be monitored via MRTG, for example. Tried this.
> 
> This traffic enters a GigE port and exit's internet-bound.
> Customer-bound return traffic comes back on same interface, all very
> complicated. By MRTG'ing Tunnel Endpoints, we maybe could have
> subtracted this from incoming and so on....
> 
> 1.	Would it be possible to terminate VPDN tunnels on a separate
> sub-interface of that same GigE port instead of loopbacks? MRTG can
> distinguish between sub-int's, right?
> 2.	By doing this, would it introduce any downsides. Loopbacks
> always up, but only one way in/out?
>
> Any thoughts appreciated.

What's wrong with monitoring the VPDN MIB (i.e. the data shown by "show
vpdn tunnel packet") which shows the L2TP traffic?

The solution with a 2nd sub-interface could work as well, you could
maybe use a VRF on the vtemplate and on the "user-vlan" to make routing
to/from the user's easier and terminate the VPDN/L2TP traffic in the
global table over the "l2tp-vlan".. On the neighboring routerhe , you
just route the address pool to the user vlan and the rest (L2TP
loopback) goes over the "lt2tp-vlan" via your regular IGP..

	oli